School selectivity and excellence: The effect of assignment to an academically selective school in Chile

• In Event: AERA Roundtable Session Saturday 9:50 am Four Seasons Ballroom 4
  In Roundtable Session: Integration Efforts in School Choice Systems (Table 12)

Abstract

In 2017 Chile began a phased implementation of a centralized admission system based on the Gale and Shapely mechanism for all public and private subsidized schools. We examine a select set of “academically excellent” (AE) schools that are permitted to phase in lottery admissions while maintaining selectivity. They utilize a dual admission system with an increasing percentage of students admitted by lottery and the remainder assigned via admission test scores. This specific policy design provides a unique opportunity to analyze the effect of assignment via lottery to a school that remains academically selective for the majority of the attendees.

It is difficult to isolate the effect of assignment to academically selective schools, as the students admitted differ from their peers not admitted due to their previous academic performance, a significant predictor of future performance. Researchers have attempted to deal with this endogeneity in descriptive studies with mixed findings through the use of controls in OLS regression, propensity score matching, and multi-level modeling (Byun, 2010; Davis, 2012; Gorard y Siddiqui, 2016; Harris y Rose, 2013; Jenkins, Micklewright y Schnepf, 2008; Kim & Lee, 2002; Levacic y Marsh, 2007; Melguizo, 2008; Quaresma, 2016; Quaresma y Orellana, 2016; Sullivan et al, 2014). Others have employed econometric techniques to attempt to isolate the effect of selective schools using differences-in-differences analysis (Rao, 2013) where time variant changes could shape the results. Others have used a regression discontinuity approach isolating the effect on students whose test scores are around the cutoff (Bucarey et al, 2014; Pop-Eleches & Urquiola, 2013), yet the results of these studies cannot be extrapolated to students far from the cutoff. This paper takes advantage of the policy in Chile to analyze the effect of assignment to AEs regardless of propensity for assignment via test scores.

We use student level administrative records from 2017-2022 to estimate the effect of assignment via lottery to AE schools. We examine the effect of assignment on multiple outcomes: enrollment at the assigned school, persistence to the end of the year in the same school, re-enrollment in the same school in the subsequent year, junior year standardized test scores, on-time graduation rate, final class rank, and college entrance exam performance. We perform intent-to-treat and treatment-of-the-treated analysis.

Students who are admitted based on their test score performance are demographically distinct from those admitted by lottery (with more low-income students admitted by lottery), but the analysis was focused only on students for whom the lottery was important in deciding their chance of admission. Results indicate that lottery winners have lower rates of attendance at the assigned school the following year than lottery losers, lower standardized test scores in their junior year exams, and have no statistically significant differences in college application, college test score, or college attendance rates. The results of the study will inform the scholarly and political debates regarding the effect of academically selective schools on student behavior, performance, and persistence.

Author

• Ngaire Noelle Honey, Universidad de O’Higgins